In this work, we developed a machine learning method to classify senescent cells via single-cell RNA sequencing data.
My contributions focused on conceptual development and coding troubleshooting.
This gave me experience working with RNA-sequencing data, including using Seurat objects and monocle trajectory analysis.
The work demonstrates that 2D in vitro models of senescence poorly recapitulate in vivo conditions. However, using complex 3D cellular models combined with sophisticated sc-RNAseq analysis provides a more reliable means of detecting senescence. The work also demonstrates the value of high-dimensional phenotype assessment as opposed to reliance on a smaller number of limited “hallmarks”.
ML Techniques: Supervised ML, Dimensionality reduction (TNSE / UMAP), PCA, Multiple Discriminant Analysis, Support Vector Machines, k-means clustering , Generalised Linear Model
